EU deepfake audio label: repeat the disclaimer in a long voiceover

Tech Policy Press says the EU Code of Practice calls for audible disclaimers on audio-only deepfakes, repeated for longer content. With Sume you can synthesise one disclaimer line, then join it between voiceover segments using POST /v1/timeline-1.0/audio with operation: "concat". The join is sample-domain, with no re-synthesis. This is a production recipe, not legal advice.

What the article lists

The summary says disclosure must happen at the latest at the moment of first exposure. For audio it lists audible disclaimers, repeated for longer content. For multimodal content it lists a visible icon that does not need user interaction. It says deployers who create realistic synthetic content must disclose it, with an exemption for private individuals using AI content for personal, non-professional use.

The page does not say how often to repeat. Pick an interval with your counsel and write it down.

Join a disclaimer into the voiceover

Concat takes 1 to 20 ordered parts, each { url, source_in?, duration? }. All URLs must be your workspace's media.sume.com audio, import them first, and all parts need the same channel layout. This request puts a disclaimer at the start, between two segments, and at the end.

curl -X POST https://api.sume.com/v1/timeline-1.0/audio \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: audio-disclaimer-001" \
  -d '{
    "operation": "concat",
    "parts": [
      { "url": "https://media.sume.com/artifacts/artf_demo/disclaimer.wav" },
      { "url": "https://media.sume.com/artifacts/artf_demo/part1.wav" },
      { "url": "https://media.sume.com/artifacts/artf_demo/disclaimer.wav" },
      { "url": "https://media.sume.com/artifacts/artf_demo/part2.wav" },
      { "url": "https://media.sume.com/artifacts/artf_demo/disclaimer.wav" }
    ]
  }'

Why concat suits this

The result is one durable file with an audio_url, duration_seconds and segments[] giving each part's start. The disclaimer is identical every time, because the same file is reused. Timeline audio costs $0.01 per job.

If you later need the whole voiceover as a lip-sync input, remember the window: H3 Max lip sync takes 5 to 14.8 seconds, so a long voiceover is split first. See inserting an ad read with timeline audio for the same concat pattern.